Overclocking out of control...

I always have some fun pushing things to the extreme, and when it comes to overclocking, I've had my very good share of voltmods and watercooling. I never went to the extreme of using stuff like liquid nitrogen or dry ice, but I void my warranties and, unfortunately, I killed a considerable amount of my hardware in the process.

I couldn't help myself, I did it again. The victim was an hand picked AMD Athlon 64 X2 5200+ F3 revision "Windsor" core, the good ones, 90nm, with the full 2MiB of L2 cache and the 65W Energy Efficient version; I am, therefore, very sad with this picture:


In my defence, I had done this a good amount of times before, never damaging any of the Athlons I tampered and obtained excellent temperature drops. This was done almost a year ago and I hadn't find any record of anyone removing an IHS on these cores.

When I picked it up, I was with too little patience and brute-forced my way in. The IHS was having a tough time coming off and, strangely, I kept being able to cut stuff inside, which I thought was the heatspreader glue. It was the resistors and that's what killed it - the core itself doesn't seem damaged, if I was able to clean it - the silver stuff is not completely solid but not your typical TIM either, seems more like some kind of alloy. You probably can't notice the detail on the picture but this "TIM" has some small air bubbles, not anything near troubling, just worth mentioning.
Both the core and the IHS have a kind of golden layer that seems to be what "solders" with this metallic stuff that goes in the middle of both, not something that I've seen before, not even from Intel.

I had a 5ºC difference between the two cores, so I thought that it might have been of bad IHS interface with the core. I had a socket 939 X2 3800+ to which I removed the IHS with very good results, in a very easy way. I should have noticed that I was cutting resistors and not glue, but since they usually are quite well soldered, I thought I was just cutting glue. The new lead free solder is also used in the new CPUs and it is usually more fragile than the lead carrying solder, certainly the solder type used with the old 939 cores. That must have been the cause of the fragility I experienced here - I never used carefulness before. If you look carefully at the picture, you'll see the Xacto knife I used cut very easily through the solder.

On a final note, this CPU was tested to run at 3.3GHz, at 1.4v, so I needed good thermals. Lukily I had another equal sample around, which overclocked a bit less. Not much to do now, other than to dwell at the interesting(but sad) find that it was.

As I said before, this was overclocking out of control, this won't surely be worth the effort or the risk. You'll have better luck by lapping the CPU since they usually aren't very flat when they come from the factory.

If any doubt persists, the Phenom is also using this new manufacturing technique.